Effect of Word-length Precision on the Performance of MIMO Systems

In this paper, we investigate the effect of quantization noise and roundoff errors involved in finite-precision (FP) signal processing on the performance of multiple-input multiple-output (MIMO) receivers under flat-fading channel conditions. We simulated the performance degradation with FP for six transmission schemes, namely the Alamouti transmit diversity scheme, spatial multiplexing (SM) under maximum-likelihood (ML) detection, SM under ordered successive interference cancellation (OSIC) for (2 times 2) MIMO, orthogonal space-time block code (STBC), quasi-orthogonal STBC for (4 times 2) MIMO and interference cancellation with two (2 times 2) Alamouti scheme. We also provide a procedure to analyze roundoff errors in FP MIMO receivers and quantify the effective decision-point SNR. The quantified variance of accumulated quantization noise and decision-point SNR for FP Alamouti scheme corroborates the simulation result. Finally, we provide the minimum word-length precision requirements for these schemes.